Optical imaging of breast tissues with near infrared (NIR) light has the potential to become an important tool in the diagnosis of breast cancer and the monitoring of deep tissue photodynamic breast cancer therapies. Unlike x-ray mammography, optical techniques are non-ionizing and may provide information about malignant tumors, although the resolution of optical mammograms is complicated by the diffuse propagation of light through tissue. Current academic and commercial approaches to optical mammography have largely focused on 2-D imaging techniques, due to the computationally intensive inverse methods employed. A powerful new method for 3-D optical imaging of breast tissues that dramatically differs from existing approaches is proposed. Recently invented techniques for both data collection and data inversion are integrated and co-optimized for use in a novel 3-D "brassiere cup" geometry that is non-compressive, comfortable, and maximizes signal to noise of the measurements. Frequently domain photon migration data will be collected at the surface of breast-shaped tissue-mimicking phantoms, with and without exogeneously introduced fluorescent contrast agents. Optical property maps for absorption, fluorescent-enhanced absorption, and fluorescence lifetime will be estimated using the rapid Bayesian inversion method known as APPRIZE. APPRIZE, unlike other inversion methods, explicitly accounts for measurement noise and system noise, and yields minimum variance estimates of optical property maps and their uncertainties. The use of innovative data-driven zonation dramatically improves the computational efficiency, stability, and accuracy of APPRIZE, relative to other competing inverse methods, and 3-D optical inversion has already been demonstrated. The continued development of sophisticated domain-decomposition strategies, along with other algorithmic improvements to APPRIZE, is expected to render high- resolution 3-D optical mammography a reality.